Our most intelligent workhorse model yet for coding and agents has arrived ⚡ Meet Gemini 3.7 Flash. — Crush that seemingly endless to-do list. Gemini Spark in the Google Gemini now uses 3.7 Flash. The new model can equip your personal AI agent to work even smarter for you... by seamlessly handling complex, multi-step tasks across your Google Workspace apps like Gmail, Google Calendar and Google Docs — Enjoy a smoother build experience. The model thinks more diligently, putting more effort into multi-step planning and tool calls. A more disciplined execution means less manual oversight and fewer retries across engineering workflows — Build more, spend less. 3.7 Flash is available through the end of the year with an introductory price of half the original 3.6 Flash cost per million tokens ($0.75/1M input tokens and $3.75/1M output tokens)show more

— Google AI Pro and Ultra subscribers can experience 3.7 Flash today via Spark in the @GeminiApp — Access the model in the Gemini Enterprise Agent Platform and Gemini Enterprise app — Build in the Gemini API via @googleaistudio and @androidstudio, and explore agent-first workflows in @antigravity — Learn more in the blog ↓

@GeminiApp Cheaper intelligence becomes much more interesting when agents can actually act on it. Better planning, tool use, and lower inference costs could push agents from assistants toward real economic actors.

@GeminiApp The more capable and affordable agents become, the closer we get to AI that can actually run workflows end to end.

@GeminiApp My gf and I tried to use Voice mode earlier today and it was so incredibly bad. Definitely needs an update. Even regular 3.0 was better. We asked it to translate Chinese/English and it randomly spoke Korean and also just said a lot of nonsense.

@GeminiApp Now that Sergey is back in a more extended capacity - we hope to see Google going back to the frontier race

@GeminiApp All based and trained on immense amounts of theft…

@GeminiApp This can be perfect workhorse model, And usage limits are good to in ai studio or use antigravity This is a perfect model for analysis and large document search. Or just quick easy prototypes.

@GeminiApp so the agents can do my to-do list while i scroll twitter? finally.

@GeminiApp DeepSWE: 48.6% → 65.3% with $0.75/M input. Hmmm interesting

@GeminiApp Gemini 3.7 Flash sounds like a notable step for AI‑assisted coding; it'll be interesting to see how its speed and agent capabilities compare to existing tools.

@GeminiApp developers are eating good with this one, love it!

@GeminiApp Most agent failures are product failures, not model failures. Scope, permissions, and handoff rules decide more than the benchmark score.

Gemini 3.7 Flash represents a significant step forward in optimizing workflows for coding and complex agent tasks. Integrating this model into the Gemini app suggests a focus on enhancing personal AI productivity. How do you see this specific iteration impacting the speed of agent-led development?

@GeminiApp AI agents are getting smarter (Gemini 3.7 Flash), now they need a clean chain to run on. Molum — L3 for AI agents, community-owned, no VC. The pack is building. $MOLUM

@GeminiApp You didn't even mention the best app it works well with. Google keep

@GeminiApp marketing team did its job well

@GeminiApp “Smartest work model” will be decided outside the launch demo. Give every model the same repo, tools and budget; then report accepted patches, retries, broken tests and human corrections. Speed matters, but cost per verified outcome is the number agents are missing.

@GeminiApp cool. now show tool logs and the invoice.

@GeminiApp mate, cost is what matters now!! Heard of DeepSeek?

@GeminiApp Gemini 3.7 Flash? Sounds like my new sidekick, but can it beat my procrastination?

@GeminiApp Gemini 3.7 Flash sounds promising for coding assistants, but real‑world performance will depend on benchmarks and integration details that aren’t public yet.

@GeminiApp 任务清单能不能少，得先看它会不会自己加活。

@GeminiApp gemini 3.7 flash looking like a solid workhorse for agents

@GeminiApp Better planning and tool use matter more than raw benchmark gains when agents are handling real workflows.

@GeminiApp Claiming 'most intelligent' without benchmarks is just marketing. How do you measure intelligence in coding-through raw speed, correctness, or adaptability?

@GeminiApp 3.7 flash is a big jump, wonder how that's gonna play out in coding tasks

@GeminiApp gemini 3.7 flash speed matters less when repo-wide edits still need regression tests

@GeminiApp Impressive progress on agentic AI for real-world workflows, multi-step task handling with less oversight is exactly where the industry's heading.

@GeminiApp Exciting advancements! AI models like Gemini 3.7 Flash can really enhance productivity, especially in automating tasks. When integrating such tools, it's essential to consider how they can optimize specific workflows, like content creation and SEO, to get the most out of them.

@GeminiApp Paying for separate developer subscriptions every time a new version drops is getting old when Magica just bundles everything for $15.

@GeminiApp Faster coding is useful. Fewer broken handoffs during long agent tasks would be the bigger win.

@GeminiApp The real bottleneck for coding agents is token density in tool output. Unless the model is actually reasoning over compressed bash history or structured diffs, it just burns context on raw command noise. Ship the agent, but optimize the environment.

@GeminiApp Our most intelligent workhorse model yet

@GeminiApp Nice